Together ZiuZ and science help patients and doctors
Image recognition. It's what ZiuZ in Gorredijk is specialised in. It started with the analysis of video material to detect child abuse, but now the company is moving more and more towards the medical sector. Projects in collaboration with the University of Groningen and the Leeuwarden Medical Centre (MCL) look very promising.
In a nutshell: you have sophisticated technology and software to analyse images, and you wonder what else you can do with it. That’s what the owner of ZiuZ Gerrit Baarda and his team have been doing in recent years. The results don’t lie. The tech company in Gorredijk is working on various projects that will go a long way in helping the medical world, and therefore also patients.
“I actually happened to hear about a problem with hyperkinetic disorders,” says Baarda. “It means that people move uncontrollably, just think of spasms. Professor Marina de Koning of the UMCG told me that in this area one in three diagnoses is not properly made. So that’s the kind of challenge we like, to see if there’s anything we can do about it.”
Nemo, as the project is called, is taking a big leap forward. A movement scientist has recently spent half of her time at the UMCG and the other half at ZiuZ; to increase the level of knowledge on both sides. Baarda: “We know how to analyse images with the aid of artificial intelligence and machine learning. But in order to succeed, we also want to know everything about hyperkinetic disorders.”
The consultation hours of specialised doctors such as Marina de Koning mostly take place in the form of video calls. This is the beginning of the diagnosis. What do you see? And can software help you look more effectively? Baarda thinks so. “We’re still in the early stages. In order for our computer to make a proper diagnosis, we need a lot of images of such disorders. Is a finger slightly out of position? Is a wrist turning differently? You have to take everything into account for the software to learn whether what it sees is a kinetic disorder.” The gathering of images and information has begun.
Another project with much potential has been named Polar. The aim is to help doctors analyse polyps on the large intestine. Surgeons now do this with the help of tiny cameras that take images from within the intestine itself. “The MCL doctor told me that given his years of experience he can almost always tell immediately whether it’s cancer or not. For doctors who are just starting out, it’s a different situation. The result is that all polyps are always removed, and it’s only once the pathologist’s work is done that it becomes clear how malignant they were.”
Image recognition from ZiuZ can help doctors determine whether surgery is immediately required. We are currently in the process of creating an image database of such polyps, together with ten hospitals now, and soon even more in the Netherlands and abroad. We then teach our software what is cancerous and what is not. It gives the surgeon an accurate tool during the operation to decide what must be removed and what can do no harm. Imagine the unnecessary cutting this saves with all the complications that come with it.”
Gerrit Baarda is more than pleased with this kind of cooperation with knowledge institutions. “We are talking to more medical specialists about projects. For us, these cooperatives mean that we are directly involved in deploying our technology in the interests of patients. That’s exactly what we’re looking for, it’s what propels us. Apart from that, a whole new, interesting world is opening up. With ethical committees, legal aspects, privacy, you name it. Incredibly interesting.”
The company in Gorredijk has expanded thanks to machine learning software that is able to recognise child abuse in images. This meant a breakthrough for law enforcement agencies, which were able to analyse much more material than they could ever do using manpower. Departments in dozens of countries are now working with this Frisian technology. The movement towards the medical sector began when Baarda, with the aid of the same image recognition, developed a machine that analyses bags of medicines. The machines that fill these bags for pharmacists make a mistake at a rate of one in two hundred. The wrong pill ends up in a bag, with potentially serious consequences for a patient. The ZiuZ machine reduced the margin of error to one in 1.3 million. The machines are now used worldwide.